The selection and location of optimal luminaries is a central aspect of architectural design. Its complexity arises due to the diversity of existing luminaires, and the problems related to the need of achieving a set of lighting goals and constraints. The use of computer simulation software can bring an improved support in decision making at design time. CAD applications for illumination assessment are generally based on a working forward strategy, where the designer selects all the design elements, in order to calculate the resulting illumination.
In this work we present an inverse approach for the selection of luminaires, where the designer defines a set of lighting intentions to satisfy, and then an optimization algorithm iterates, converging to a feasible and optimal solution. The method allows to use a database consisting of hundreds of luminaires and a set of possible locations. In each iteration, after the first reflection of a potential configuration is calculated, the radiosity method is used to compute the final illumination of the scene.